An attempt to convert Kotlin Lingua to Java Lingua

[azagniotov]

Context

@pemistahl Peter,

After coming across the:

• Replace native implementation with Java bindings to WASM module #214 and the Rewrite Lingua in Java #188

I decided to give it a go and lend a hand, or two. I do realize that without knowing your own progress with the #214, my effort may be a futile effort. But, I still decided to jump in.

I understand that you may be busy with real life and a young family, so whenever you have some feedback, I will really appreciate it. At this point, I would like to ask you how is it looking so far. Also, please let me know if you have any concerns. In any way, should there be any issues where I should not proceed further, I am happy to close the PR.

Thank you.

The State of Things

What has been migrated/achieved

• build.gradle has been migrated in order to build the project and run the tests
• The commits on the PR are atomic enough, so I hope this will enable you to review this with more ease.
• All the migrated test cases pass (apart from what has been outlined in Concerns so far below)
• Gradle plugin 'com.github.sherter.google-java-format' was added to help with formatting, e.g.: ./gradlew googleJavaFormat

What has not been migrated/achieved

• build.gradle artifact packaging, reporting and publishing still need to be tested out
• api package: LanguageDetector and LanguageDetectorBuilder, plus their respective tests
• accuracyReport package classes
• internal.io package classes

Concerns so far

• The test TrainingDataLanguageModelTest.assertThatUnigramLanguageModelIsCorrectlySerializedToJson is failing because the obtained JSON object keys are not in the same order as the expected JSON. Therefore JSON object string comparison fails.

[pemistahl]

Hi, thank you for your effort in this matter. But, I'm afraid to say, your effort was in vain, I'm sorry. I won't improve the JVM implementation anymore in favor of using Java bindings to the WASM module of the Rust implementation of the library, as you have already read yourself. So I'm closing this PR now. My approach will lead to better runtime performance and much less memory consumption, especially if I add more languages in the future. So this is the way to go.

[azagniotov]

@pemistahl Peter, thank you and not a problem at all. I suspected as much.

Would you be able to provide a rough timeline when would it be possible to use the Rust implementation via the Java bindings/WASM? If this already possible, do you have a compilation/usage guide?

[pemistahl]

Would you be able to provide a rough timeline when would it be possible to use the Rust implementation via the Java bindings/WASM?

No, not yet. My spare time has never been as limited as it is now. So I cannot give you a timeline right now.

[azagniotov]

@pemistahl Peter, understood. We all have real lives.. and I also have a young family :) Just to trouble you with one last question on this topic - will the future artifact (i.e.: the Rust implementation via the Java bindings/WASM) be available for a download from Maven Central?

[pemistahl]

Yes, the new release will be available at Maven Central. It will look and behave like a normal Java library.